Sensitivity and Specificity of the Hydrogen Breath ... - Semantic Scholar

5. Schellenkens APM, Dander GTB, Thornton W, Van Geoenstein T. Sources of variation in the column chromatographic determination of HbA1. Clin Chem 27, 94-99 (1981). 6. Dix D, Cohen P, Kingsley S. Interference by lactescence in glycohemoglobin analysis. Clin Chem 25, 494-495 (1979). 7. Fluckiger R, Winterhalter JH. In vitro synthesis of hemoglobin A1,. FEBS Left 7,356-360 (1976). 8. Aleyassine H, Gardiner RJ, Blankstein LA, Dempsey ME. Agar gel electrophoretic determination of glycosylated hemoglobin: Effect of variant hemoglobins, hyperlipidemia, and temperature. Clin Chem 28, 474-477 (1981). 9. Klenk DC, Hermanson GT, Krohn RJ, et al. Determination of glycosylated hemoglobin by affinity chromatography: Comparison with colorimetric and ion-exchange methods, and effects of common

interferences. Clin Chem 25, 2088-2094 (1982). 10. Yatscoff RW, Braidwood JL. Comparison of column chromatographic, colorimetric and electrophoretic methods for determination of glycosylated hemoglobin (HbA1). Clin Biochem, in press (1982). 11. Goldstein DE, Peth DB, England JD, et al. The effect of acute changes in blood glucose on HbA,. Diabetes 29, 623-628 (1980). 12. Aleyassine H. Low proportions of glycosylated hemoglobin associated with hemoglobin S and hemoglobin C. Cliii Chem 25, 1484-1486 (1979). 13. Nathan DM, Avezzano BJ, Palmer JC. A rapid method for eliminating labile glycosylated hemoglobin. Clin Chem 23,512-513 (1982). 14. Pembrey ME, Weatherall DJ, Clegg JB. Maternal synthesis of hemoglobin F in pregnancy. Lancet i, 1350-1354 (1973).

CLIN. CHEM. 29/3, 545-548 (1983)

Sensitivity and Specificity of the Hydrogen Breath-Analysis Test for Detecting Malabsorption of Physiological Doses of Lactose Jorge Luis Rosado and Noel W. Solomons’ We examined the changes in sensitivity and specificity that would occur with alterations in the sample-collection schedule and (or) cutoff criterion for the increase in hydrogen concentration in breath after administration of doses of lactose in the dietary range. In a breath-analysis test to classify individuals as lactose-absorbers or lactose-malab-

sorbers, 41 subjects

drank

360

mL of intact

cow’s

milk,

containing 18 g of lactose, and breath samples were collected and analyzed at 30-mm intervals for 5 h. An increase in H2 concentration of 20 UL above basal values at any of the 10 intervals was diagnostic of malabsorption. Increases of 18 or 15 UL were only 85% as specific in classifying the same individuals. Reduction in the number of samples tested per subject uniformly reduced the sensitivity. However, a simplified procedure suitable for field studies (in which four samples-at 0, 2, 3, and 4 h-are collected and analyzed with ‘20 L/L as the cutoff value) gives 80% sensitivity and 100% specificity, as compared with the 11-sample procedure.

AdditionalKeyphrases:lactose malabsorption ing

screen-

cutoff value

The principle that the colonic fermentation of incompletely absorbed carbohydrates results in increased excretion of breath hydrogen (H2) has led to the development of nomnvasive procedures for diagnosing carbohydrate malabsorption. As originally applied to the diagnosis of lactose malabsorption, the conventional dose of substrate was 50 g of lactose (1-4). However, because the technique is sufficiently sensitive to detect the malabsorption of as little as 2 g of Department of Nutrition and Food Science, Massachusetts Institute of Technology, Cambridge, MA 02139. 1Address correspondence to this author, International Nutrition Program, Massachusetts Institute of Technology, Em 20B-213, 18 Vassar St., Cambridge, MA 02139. Received Sept. 27, 1982; accepted Dec. 8, 1982.

incompletely absorbed carbohydrate (5), and because questions regarding carbohydrate tolerance usually involve dietary issues, the use of physiological doses of carbohydrate in their natural food forms has been advocated (6, 7). Breath excretion of H2 can be monitored by a continuous, closed collection system (8) or by determining the concentration of H2 in expired air at fixed intervals (1, 3, 4, 9). With the latter approach, the criterion for determining malabsorption is usually based on the size of the increase from the fasting, basal concentrations of H2. An increase of ‘20 jZL/L 2 h after an aqueous, pharmacological dose of lactose is a common cutoff value (3,4). This simple approach, requiring only a single pre-dose and a single post-dose collection, has a sensitivity and specificity of 100% as an index of lactase deficiency in adult subjects (3). As the administered dose of substrate is decreased into the range of quantities encountered physiologically, the magnitude of breath H2 production will also decrease. Thus, the frequency and timing of breath collections becomes an important determinant of the sensitivity and specificity of H2 breath tests involving physiological amounts of lactose. This is particularly important in field studies, in which it is desirable to limit both the number of collection intervals and the time that the subjects must remain fasting. We have analyzed the results from a series of 41 adult subjects receiving 18 g of lactose from whole milk and report here our findings and conclusions.

Materials and Methods The subjects were 41 healthy adults who participated in a screening test to select volunteers for a study to evaluate the effectiveness of in vivo hydrolysis of lactose (ma. submitted for publication). The protocol was approved by the MIT Committee on the Use of Humans as Experimental Subjects, and all subjects signed consent forms after the nature, purposes, and risks of the test had been explained. Mixed expired air was collected from fasting subjects by having them breathe normally through a low-resistance two-way Hans Rudolph valve connected to a 5-L anesthesia gas-bag with nipple. A sample of the breath was aspirated CLINICALCHEMISTRY, Vol. 29, No. 3, 1983 545

into a 60-mL plastic syringe fitted with a three-way stopcock. The air was maintained in the syringe until analysis within 8 h. The subject subsequently drank 360 mL of intact cow’s milk, containing 18 g of lactose. The breath-sampling procedure was repeated at 30-mm intervals over the next 5 h. During this time, the subjects were ambulatory and were allowed to drink water, but not to smoke or eat. Breath H2 concentration was measured with a MicroLyzer (Quintron Instruments, Co., Milwaukee, WI), a thermal conductivity detector-based gas-solid chromatographic system in which aspirated room-air is the propulsion gas (10). Once the instrument is calibrated with a compressedgas standard of known H2 concentration (Scotty Gas II; Supelco, Bellefonte, PA), the peak height of an unknown breath sample is read from a digital panel meter. We arbitrarily interpreted the tests as “positive,” that is, as signifying a biologically significant failure to complete lactose absorption, when the maximum increase above the fasting, basal breath H2 concentration was ?‘25 41L at any of the 30-mm intervals after the milk was drunk. This more stringent criterion, 5 .tUL higher than the conventional cutoff criterion (3, 4), was chosen to provide for a clear separation of malabsorbers for inclusion into the longitudinal study of in vivo hydrolysis by exogenous lactases (ms. submitted for publication). Sensitivity, an estimate of the efficiency with which a diagnostic test obtains truly positive results, is defined as the number of positive tests divided by the number of individuals affected, e.g., with lactose malabsorption. Specificity, an estimate of the efficiency with which truly negative results are obtained, is defined here as the number of negative tests divided by the number of individuals without lactose malabsorption. As a starting point, we have defined the results of the complete 11-sample standard collection procedure as the “true” diagnosis.

50

40

2

0 R

30

zw C.)

20 C.) 04

I I

I-

10

2

-10

0

1

2

3

4

5

TIME IN HOURS

Fig. 1. Mean post-ingestionchanges in breath H2 concentration (L/L) after ingestion of 360 mLof intact milk(18 g of lactose) by21 lactoseabsorbers (LA) (lower curve) and20 lactose-malabsorbers (LM) (upper cwve) serially over5 h The shadedareas bracketing the curves represent ± 1 SDof the meansfor each 30-mm sampling interval for the respective cohorts: LM (vertical sthpes); LA (dotted area). See text for diagnostic critena

Results Diagnosis of lactose absorption status. Of the 41 subjects, 20 had an increase in breath H2 concentration of 25 UL at some interval during the 5-h collection period, while the maximum increase of the others was 03. 2 U-

0

55

55

LU

administered 2

68

29

40

30

29 -

25

28

31

29

0-

//

// 1

‘/7 /7

2

3 TIME

45

29

25

81

1/ 7/

4

carbohydrate

is represented

by an increment

in breath H2 concentration of ‘10 UL. Had this criterion been applied to our sample of 41 individuals, 24 would have been classified as malabsorbers and 17 as absorbers. However, a distinction must be made between “incomplete absorption” and “biologically significant malabsorption.” Whether

5

IN HOURS

Fig. 2. Frequencydistribution of the time of the maximum increase in breathH2 concentration over the ten 30-mm samplingintervalsfor each of the 20 indMdualsclassified as LMs The number inside the rectangle represents the magnitude of the maximum breath H2increments(pJiL)

ly popular over the past decade. As originally applied to the diagnosis of lactose malabsorption, the conventional, pharrnacological dosages of carbohydrate, 2 g/kg of body weight up to a total dose of 50 g, derived from traditional “lactose tolerance tests,” were used (1-4). Results of the H2 breath test after this dose of lactose were found to be the most reliable indirect index of lactase deficiency (3); however, the original basis for the use of pharmacological doses of lactose was to allow the detection and discrimination of increments in blood glucose sufficient to separate LM from LA on the basis of post-lactose glycemia. The more responsive breath H2 analysis-as little as 2 g of carbohydrate not absorbed provides a detectable increase in the excretion of breath H2-permitted the application of oral doses of lactose in the physiological (dietary) range (7, 10). In recent years, the trend has been toward the use of the amount of lactose in 8to 12-oz. (240- to 360-mL) glasses of whole milk, i.e., doses of 12 to 18 g of lactose (6, 12-21). The other important trend is the use of the H2 breath test in field studies and population surveys (18, 22-25). In a clinical setting, sampling breath for H2 determinations as frequently as one to six times an hour is not unreasonable, but in population studies, a more limited sampling regimen must be imposed. With the 50-g dose of lactose, a pre-dose and a 2-h post-dose collection were the only samples obmined by Newcomer et al. (3) and were sufficiently sensitive. We have observed occasionally, however, that even at this high a dose in aqueous solution, the increase of breath H2 by ‘20 LIL does not occur until the third hour post-dose or later (6). Thus, we were concerned for the potential loss of diagnostic sensitivity with reduced dosages of lactose, espe-

that escapes small-intestinal removal and reaches the colon to produce symptoms and evolve H2 derives from an original oral dose of 10 g or 50 g of lactose, it seems reasonable that the same absolute amount of incompletely absorbed substrate would be required to cause overt manifestations of intolerance in either situation. In an earlier study involving a dose of 12.5 g of lactose, an increase in breath 112 concentrations of at least 20 1.dJL was generally required before a subject reported experiencing symptoms (6). The ± 1 SD limits around the mean of the interval changes in breath H2 concentration in our present subjects (Figure 1) similarly indicate the limitations of using small increases in breath H2 concentration as cutoff values. The later appearance of the peak increase in breath H2 concentration after drinking milk lactose than after aqueous lactose is consistent with a slower rate of gastric emptying of the food vehicle (28, 29), so that peak increments occur in the later hours of a 5-h collection period. Using only the results of the analyses of the breath samples from the third and fourth hour post-dose gave a test sensitivity of 75%. Although adding one additional collection improves the sensitivity slightly, including results from the fifth hour prolongs the overall period of fasting while increasing sensitivity by only 5%. This same increase in sensitivity can be achieved by measuring the 2-h post-dose expired air, thus decreasing the number of breath collections from 11 to four, but still achieving an 80% sensitivity and 100% specificity with a physiological dose of lactose. In a previous report in which we measured breath H2 at 30-mm intervals in conjunction with a physiological (12.5 g) dose of lactose (6), we found the same (100%) sensitivity for samples obtained “on the hour” or “on the half hour” during a 6-h period. In the present series, using only half of the post-ingestion breath samples gave only 90% sensitivity. This underscores the need for prior determination of the effects on the diagnostic accuracy of breath test procedures when modifications in cutoff criteria or sampling intervals the carbohydrate

are introduced.

In summary, a reasonable approach for field studies involving physiological amounts of lactose, one that would have 80% sensitivity and 100% specificity, would involve the following criterion: an increase of ‘20 uL’L in any one of the sampling intervals at 2, 3, and 4 h after ingestion of 360 mL (12 oz.) of whole milk. Examination of dietary issues CLINICALCHEMISTRY, Vol. 29, No. 3, 1983 547

in lactose utilization is best performed with physiological amounts of carbohydrate (7), but a careful evaluation of the sensitivity and specificity of the diagnostic criteria is an essential prelude to any screening test.

References 1. Calloway DH, Murphy EL, Bauer D. Determination of lactose intolerance by breath analysis. Am J Dig Dis 14, 811-815 (1969). 2. Levitt MD, Donaldson RM. Use of respiratory hydrogen (H2) excretion to detect carbohydrate malabsorption. J Lab Clin Med 75. 937-945

(1970).

3. Newcomer AD, McGill DB, Thomas PJ, Hofmann AF. Prospective comparison of indirect methods for detecting lactase deficiency. N EngI J Med 293, 1232-1236 (1975). 4. Met.z G, Jenkins DJA, Peters JJ, et a!. Breath hydrogen as a diagnostic method for hypolactasia. Lancet I, 1155-1 158 (1976). 5. Levitt, MD. Production and excretion of hydrogen gas in man. N Engi ,J Med 281, 122-127 (1969). 6. Solomons NW, Garcia-Ibanez R, Viteri FE. Hydrogen breath test of lactose absorption in adults: The application of physiological doses and whole cow’s milk sources. Am J Gun Nutr 33, 545-554 (1980).

7. Torun B, Solomons NW, Viteri FE. Lactose malabsorption and lactose intolerance: Implications for general milk consumption. Arch Latinoam Nutr 29, 445-494 (1979). 8. Bond JH, Levitt MD. Use of pulmonary hydrogen (H2) measurements to quantitate carbohydrate malabsorption: Study of partially gastrectomized patients. J Clin Invest 51, 1219-1225 (1972). 9. Solomons NW, Viteri FE, Hamilton LH. Application of a simple gas chromatographic technique for measuring breath hydrogen. J Lab Clin Med 90, 856-862 (1977). 10. Christman NT, Hamilton LH. A new chromatographic instrument for measuring trace concentrations of breath hydrogen. J Chromoiogr 229, 259-265 (1982). 11. Solomons NW. Diagnosis and screening techniques for lactose maldigestion: Advantages of the hydrogen breath test. In Lactose Digestion: Clinical and Nutritional Implications, DM Paige, TM Bayless, Eds., Johns Hopkins Univ Press, Baltimore, MD, 1981, pp

91-109. 12. Gearhart

HL, Bose DP, Smith CA, et al. Determination of lactose malabsorption by breath analysis with gas chromatography. Anal Chem 48, 393-398 (1976). 13. Bond JH, Levitt MD. Quantitative measurement of lactose absorption. Gastroenterviogy 70, 1058-1061 (1976). 14. Caskey DA, Payne-Bose D, Welsh JD, et al. Effects of age on

548

CLINICAL CHEMISTRY, Vol. 29, No. 3, 1983

lactose malabsorption

in Oklahoma Native Americans as determined by breath H2 analysis. Am J Dig Di.s 22, 113-116 (1977). 15. Payne-Bose D, Welsh JD, Gearhart HL, Morrison RD. Milk and lactose-hydrolyzed milk. Am J Gun Nutr 30, 695-697 (1977). 16. Rorick M, Scrimshaw NS. Comparative tolerance of elderly from differing ethnic backgrounds to lactose-containing and lactosefree dairy drinks: A double-blind study. J Gerontol 34, 191-196 (1979). 17. MacLean WC, Fink BB. Lactose malabsorption by premature infants: Magnitude and clinical significance. J Pediatr 97, 383-388 (1980). 18. Ellestad-Sayed JJ, Levutt MD, Bond JH. Milk intolerance in Manitoba Indian school children. Am J Clin Nutr 33, 2198-2201 (1980). 19. Brown KIT, Khatun M, Parry L, Ahmed MG. Nutritional consequences of low dose milk supplement consumed by lactosemalabsorbing children. Am J Clin Nutr 33, 1054-1063 (1980). 20. Kirschner BS, de Favaro MV, Jensen W. Lactose malabsorption in children and adolescents with inflammatory bowel disease. Gastroenterology 81, 829-832 (1981). 21. Payne DL, Welsh JD, Manion CV, et al. Effectiveness of milk products in dietary management of lactose malabsorption. Am J Gun Nutr 34, 2711-2716 (1981). 22. Newcomer AD, Thomas PJ, McGill DB, Hofmann AF. Lactase deficiency: A common genetic trait of American Indians. Gastroenterology 72, 234-237 (1977). 23. Newcomer AD, Gordon H, Thomas PJ, McGill DB. Family studies of lactase deficiency in the American Indian. Cast roe nterology 73, 985-988 (1977). 24. Newcomer AD, McGill DB, Thomas PJ, Hofmann AF. Tolerance to lactose among lactose-deficient American Indians. Castroenterology 74, 44-46 (1978). 25. Brown KIT, Parry L, Khatun M, Ahmed MG. Lactose malabsorption in Bangladeshi village children: Relation with age, history of recent diarrhea, nutritional status and breast feeding. Am J Gun Nutr 32, 1962-1969 (1979). 26. Barr RG. Limitations of the hydrogen breath test and other techniques for predicting incomplete lactose absorption. In Lactose Digestion: Clinical and Nutritional implications (see ref. 11), pp 110-114. 27. Bayless TM. Lactose malabsorption, milk intolerance and symptom awareness in adults. Ibid, pp 117-123. 28. Solomons NW, GarcIa-Ibanez R, Viteri RE. Reduced rate of breath

hydrogen

young children

(H2) excretion using

with

lactose

tolerance

tests

in

whole milk. Am J Clin Nutr 32, 783-786

(1979). 29. Welsh JD, Payne DL, Manion C, et al. Interval sampling of breath hydrogen (H2) as an index of lactose malabsorption in lactase-deficient subjects. Dig Dis Sci 26, 681-685 (1981).